SnapFusion technology greatly improves the speed of AI image generation

2023-06-28 16:44:25 Author: Zhao Wusheng

Recently, the research department of the social media company announced the development of a new technology called SnapFusion, which can greatly increase the speed of AI-generated images, shortening it from several minutes or even hours to less than two seconds. SnapFusion can quickly generate high-definition images through text prompts entered by the user. By optimizing the network structure and removing noise, this technology achieves an efficient model and is able to maintain image quality. Now, users can run the model on a mobile device, generate images based on text prompts, and get results in seconds. The published research paper details SnapFusion technology and showcases artwork created with the technology. The paper proposes an efficient UNet by identifying the redundancy of the original model and reducing the computation of the image decoder through data distillation. Step distillation is further enhanced by exploring training strategies and introducing regularization of unsupervised classifiers. This new technology will still take some time to be available to a wide range of users, but it has the potential to power high-quality AI-generated imaging experiences on mobile devices in the future.

The above is the detailed content of SnapFusion technology greatly improves the speed of AI image generation. For more information, please follow other related articles on the PHP Chinese website!